基于灰度共生矩阵的沥青路面病害雷达图像纹理特征提取研究

Research on Texture Feature Extraction of Asphalt Pavement Disease Radar Image Based on Grayscale Symbiosis Matrix

随着电磁波探测在道路领域的日益推广,探地雷达逐渐成为道路内部状况诊断的有效手段。通过探地雷达设备获取的道路内部雷达图像经校零、背景去除、滤波等处理后因图像差异性较大目前没有较为统一的解释。为了探究不同参数下的灰度共生矩阵特征对不同病害图像的适用性,本文首先通过正交试验设计对灰度共生矩阵参数进行组合,然后在不同参数下利用灰度共生矩阵算法对原始雷达图像进行特征计算,最后得到不同参数设置下对比度、逆方差、能量、熵值四种特征图像。结果表明:当滑动窗口为11×11、统计方向为45°、灰度级数为8时,层间黏结不良的熵特征更为明显;滑动窗口为5×5、统计为方向90°、灰度级数为128时,裂缝的熵特征更为明显。本文提出的探地雷达图像纹理特征适用性参数设置对图像分割、识别及数据增强等有一定的参考价值。

With the increasing promotion of electromagnetic wave detection in the road field,ground penetrating radar has gradually become an effective method for road internal condition diagnosis.There is no unified interpretation of the radar images inside the road obtained by ground penetrating radar equipment after processing such as zeroing,background removal,and filtering,due to the large differences in images.In order to explore the applicability of grayscale co-existence matrix features under different parameters to different disease images,this paper first combines the gray scale co-existence matrix parameters through orthogonal experimental design,and then uses the grayscale co-existence matrix algorithm to calculate the features of the original radar image under different parameters,and finally obtains four feature images of contrast,inverse variance,energy and entropy under different parameter settings.The results show that the entropy characteristics of poor interlayer bonding are more obvious when the sliding window is 11×11,the statistical direction is 45°,and the gray level is 8.The entropy characteristics of cracks are more obvious when the sliding window is 5×5,the statistic direction is 90°,and the gray level is 128.The parameter setting of texture feature applicability of GPR image proposed in this paper has certain reference value for image segmentation,recognition and data enhancement.